Research map / Label-efficient and robust learning / Domain adaptation and generalization
Is It Necessary to Transfer Temporal Knowledge for Domain Adaptive Video Semantic Segmentation?
Xinyi Wu, Zhenyao Wu, Jin Wan, Lili Ju, Song Wang
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add temporal augmentation (with consistency learning) (ablation) | I2VDA · VIPER -> Cityscapes (ablation) | mIoU: 47.0 -> 51.2 | Improved |
| add temporal-augmented consistency learning (ablation) | I2VDA · VIPER -> Cityscapes (ablation) | mIoU: 45.6 -> 51.2 | Improved |
| add temporal augmentation alone to baseline (ablation) | Baseline · VIPER -> Cityscapes (ablation) | mIoU: 44.0 -> 45.6 | Improved |
| add occlusion handling in consistency loss (ablation) | I2VDA · VIPER -> Cityscapes (ablation) | mIoU: 49.7 -> 51.2 | Improved |
| add entropy map in consistency loss (ablation) | I2VDA · VIPER -> Cityscapes (ablation) | mIoU: 49.6 -> 51.2 | Improved |
| copy proxy network every 8k instead of 1k iterations (ablation) | I2VDA · VIPER -> Cityscapes (ablation) | mIoU: 48.9 -> 51.2 | Improved |
| launch proxy network at 8k instead of 1k iterations (ablation) | I2VDA · VIPER -> Cityscapes (ablation) | mIoU: 48.3 -> 51.2 | Improved |
| launch proxy network at 8k instead of 15k iterations (ablation) | I2VDA · VIPER -> Cityscapes (ablation) | mIoU: 50.2 -> 51.2 | Improved |
| random temporal interpolation t instead of fixed t=0.5 (ablation) | I2VDA · VIPER -> Cityscapes (ablation) | mIoU: 48.4 -> 51.2 | Improved |
| delta 0.3 instead of 0.1 (ablation) | I2VDA · VIPER -> Cityscapes (ablation) | mIoU: 48.5 -> 51.2 | Improved |
| alpha 0.5 instead of 0.9 (ablation) | I2VDA · VIPER -> Cityscapes (ablation) | mIoU: 48.8 -> 51.2 | Improved |
| eta 1.0 instead of 2.0 (ablation) | I2VDA · VIPER -> Cityscapes (ablation) | mIoU: 50.2 -> 51.2 | Improved |
Closest papers by meaning
- Pay Attention to Target: Relation-Aware Temporal Consistency for Domain Adaptive Video Semantic Segmentation (AAAI 2024 · similarity 0.91)
- Domain Adaptive Video Segmentation via Temporal Consistency Regularization (ICCV 2021 · similarity 0.91)
- Domain Adaptive Video Segmentation via Temporal Pseudo Supervision (ECCV 2022 · similarity 0.90)
- On the Road to Online Adaptation for Semantic Image Segmentation (CVPR 2022 · similarity 0.89)
- Spatio-Temporal Pixel-Level Contrastive Learning-Based Source-Free Domain Adaptation for Video Semantic Segmentation (CVPR 2023 · similarity 0.89)
- Domain Adaptive Semantic Segmentation With Self-Supervised Depth Estimation (ICCV 2021 · similarity 0.89)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
